On Thu, Jan 31, 2019 at 08:55:17AM +0100, Christoph Hellwig wrote:
> We already shortcut xfs_map_blocks for COW mappings, but there is just
> as little reason to start writeback beyond i_size in the data fork.
>
> Note that this has to be just an optimization as hole punches can unmaps
> block just like truncate, and we need to handle that case further down
> in the low-level code.
>
> Signed-off-by: Christoph Hellwig <hch@xxxxxx>
> ---
> fs/xfs/xfs_aops.c | 34 ++++++++++++++--------------------
> 1 file changed, 14 insertions(+), 20 deletions(-)
>
> diff --git a/fs/xfs/xfs_aops.c b/fs/xfs/xfs_aops.c
> index 8bfb62d8776f..9c2a1947d5dd 100644
> --- a/fs/xfs/xfs_aops.c
> +++ b/fs/xfs/xfs_aops.c
> @@ -348,6 +348,20 @@ xfs_map_blocks(
> if (XFS_FORCED_SHUTDOWN(mp))
> return -EIO;
>
> + /*
> + * If the offset is beyond the inode size, we know that we raced with
> + * truncate. No point in doing calling any lower level code, just
> + * return a hole so that the writeback code skips writeback for the
> + * rest of the file.
> + */
> + if (offset > i_size_read(inode)) {
> + wpc->imap.br_startoff = offset_fsb;
> + wpc->imap.br_blockcount = end_fsb - offset_fsb;
> + wpc->imap.br_startblock = HOLESTARTBLOCK;
> + wpc->imap.br_state = XFS_EXT_NORM;
> + return 0;
> + }
> +
The code looks fine, but I don't see any more value in this code than
the similar code down in xfs_iomap_write_allocate(). The comment implies
this skips writeback for the rest of the file, but AFACT the higher
level page->index code in xfs_do_writepage() already does that. All
these checks do is skip the remaining blocks in the current page. When
you consider that we're most likely sending an I/O in the latter case
either way, I'm curious why we'd bother to keep this around at all.
Brian
> /*
> * COW fork blocks can overlap data fork blocks even if the blocks
> * aren't shared. COW I/O always takes precedent, so we must always
> @@ -388,26 +402,6 @@ xfs_map_blocks(
> xfs_iunlock(ip, XFS_ILOCK_SHARED);
>
> wpc->fork = XFS_COW_FORK;
> -
> - /*
> - * Truncate can race with writeback since writeback doesn't
> - * take the iolock and truncate decreases the file size before
> - * it starts truncating the pages between new_size and old_size.
> - * Therefore, we can end up in the situation where writeback
> - * gets a CoW fork mapping but the truncate makes the mapping
> - * invalid and we end up in here trying to get a new mapping.
> - * bail out here so that we simply never get a valid mapping
> - * and so we drop the write altogether. The page truncation
> - * will kill the contents anyway.
> - */
> - if (offset > i_size_read(inode)) {
> - wpc->imap.br_blockcount = end_fsb - offset_fsb;
> - wpc->imap.br_startoff = offset_fsb;
> - wpc->imap.br_startblock = HOLESTARTBLOCK;
> - wpc->imap.br_state = XFS_EXT_NORM;
> - return 0;
> - }
> -
> goto allocate_blocks;
> }
>
> --
> 2.20.1
>
